We are entering a very exciting era in umbilical cord blood transplantation (UCBT), where many of the associated formidable challenges may become treatable by ex vivo graft manipulation and/or adoptive immunotherapy utilizing specific cellular products. We envisage the use of double UCBT rather than single UCBT for most patients; this allows for greater ability to treat larger patients as well as to manipulate the graft. Ex vivo expansion and/or fucosylation of one cord will achieve more rapid engraftment, minimize the period of neutropenia and also give certainty that the other cord will provide long-term engraftment/immune reconstitution. The non-expanded (and future dominant) cord could be chosen for characteristics such as better HLA matching to minimize GvHD, or larger cell counts to enable part of the unit to be utilized for the development of specific cellular therapies such as the production of virus-specificT-cells or chimeric-antigen receptor T-cells which are reviewed in this study.
Umbilical cord blood transplantation (UCBT) was first utilized to treat a 5-year-old child with severe aplastic anemia because of Fanconi Anemia in 1988, who had a UCBT from an HLA-identical sibling.1 In the pediatric setting, UCBT, when compared with matched-unrelated donor (MUD) bone marrow (BM) or mobilized PBSC transplantation, is associated with delayed engraftment, less acute and chronic GvHD and overall similar relapse rate and survival.2
Initial results in adults were poor, with 40% mortality at day 100 (D100).3 However, these have improved with better patient and cord selection and supportive care.4 Adult patients treated for AML or ALL have lower relapse rates with UCBT compared with other stem cell sources, but higher treatment-related mortality (TRM) and overall similar survival.5
Only ~30% of patients who require allogeneic stem cell transplantation (alloSCT) will have a matched sibling donor. Many patients without a matched sibling will have an adult MUD, but this is often not the case in patients from ethnic minorities.6 There is a lower incidence of acute GvHD (aGVHD) and chronic GvHD (cGvHD)after UCBT relative to transplant from an adult donor for a given degree of HLA mismatch;7 this has substantially increased the number of patients potentially able to undergo alloSCT as patients undergoing 4–6/6 matched UCBT have a lower incidence of aGvHD than an 8/8-matched sibling or adult MUD transplant and a lower incidence of cGvHD than an 8/8-matched MUD, but not sibling donor.8 The development of reduced-intensity conditioning (RIC) regimens has also extended eligibility to older patients who could not tolerate myeloablative conditioning.9
Challenges associated with UCBT
Despite the advantages of UCBT and the advances in management of patients undergoing the procedure, substantial challenges remain.
Delayed engraftment, failed engraftment and unit selection
The low numbers of hematopoietic progenitors present within a cord blood (CB) unit leads to slower engraftment than with other stem cell transplant sources and a higher risk of failed engraftment.5 Overall rates of engraftment in adults are ~80–90%.7, 10, 11, 12, 13, 14 In contrast, failure to engraft is rare in patients receiving alloSCT from an adult MUD with myeloablative conditioning (<1%) when 6/6 matched at class-I HLA alleles. However, failed engraftment does increase to 21% in the presence of a single class-I HLA antigen mismatch or multiple class-I allele/antigen mismatches.15
Improving engraftment rates: cord selection
Cell dose is the most important determinant of successful UCBT in adults.10, 11, 13, 16 UCBT with a single cord (sUCBT) containing <2.5 × 107/kg total nucleated cells is associated with poor engraftment, high TRM and poor survival.16 Double UCBT (dUCBT) compared with sUCBT may allow this cell-dose threshold to be reached for adult patients.5 dUCBT, relative to sUCBT, is associated with lower relapse rates and improved leukemia-free survival in patients with acute leukemia transplanted in CR1, at a cost of higher rates of grade II–IV, but not grade III–IV, GvHD, with no difference in TRM.17 Other factors contributing to risk of graft rejection include the presence of donor-specific anti-HLA antibodies, which have been associated with both failed engraftment and higher TRM.18 Recipient antibodies against mismatched donor HLA-DP have also been shown in 10/10 matched MUD alloSCT to be associated with a higher incidence of graft failure and mortality.19 Risk of rejection also increases with increasing numbers of mismatches at HLA-A, -B, -C and -DR loci.20
Patients receiving dUCBT often have a period of transient mixed chimerism and almost universally have eventual dominance of one cord and disappearance of the other.9, 21, 22 This is owing to CD8+ T-cell-mediated immunologic rejection of one cord unit by the other.23 No factors have been identified to predict which unit will become dominant.24
Improving rapidity of engraftment: novel strategies
Neutrophil engraftment in UCBT is delayed relative to PBSC or BM, occurring between 20–30 days when patients receive myeloablative conditioning.25 It depends strongly on cell dose.10, 11, 13, 26, 27, 28 RIC accelerates neutrophil recovery and is associated with a more rapid quantitative and qualitative T-cell recovery. However, whether this translates into improved outcomes is unclear.29
Recently, there has been considerable interest in the development of techniques for the manipulation of CB ex vivo to accelerate engraftment. These have focused on either expanding the number of progenitor cells within the unit or on modifications to stem cells to improve homing to marrow niches.
Culture of CB cells with mesenchymal stromal cells enhances the expansion of primitive and mature hematopoietic progenitor cells (HPCs).30 de Lima et al.31 demonstrated a significantly reduced time to neutrophil (15 vs 24 days) and platelet (42 vs 49 days) engraftment in patients receiving dUCBT where one cord was cultured ex vivo with mesenchymal stromal cells relative to historical control patients receiving unmanipulated dUCBT. Chimerism results showed, however, that long-term reconstitution of hematopoiesis was almost entirely derived from the unmanipulated CB unit. This suggests that the culture process expands committed myeloid and megakaryocytic progenitors in the graft capable of rapid engraftment at the expense of depleting the graft of cells capable of producing long-term marrow repopulation. In clinical use, certainty surrounding which CB unit will eventually provide long-term hematopoietic and immunological reconstitution will potentially be useful, as the unit with less ideal HLA matching could be used for expansion. Long-term engraftment by the better-matched cord could potentially reduce the risk of aGvHD and cGvHD. Alternatively, if there were no specific immunologic advantages to one cord over another, one cord could be utilized for cellular therapy procedures such as generation of viral-specific T-cells (see below); these could then be expected to persist long-term in the absence of competition from the other CB unit, which would be expanded ex vivo to provide rapid hematopoietic recovery.
An alternative strategy for ex vivo expansion of HPCs utilizes culture in the presence of NOTCH ligand. NOTCH is expressed in CD34+ HPCs32 and retroviral transduction of a constitutively expressed form immortalizes HPCs.33 Incubation of CB progenitors with NOTCH-ligand produced a 100-fold increase in CD34 numbers and enhanced repopulating ability in an immunodeficient mouse model.34, 35. Preliminary results in a phase-I human study have shown more rapid neutrophil engraftment (16 vs 24 days in a similarly and simultaneously treated cohort). Similarly to the results using mesenchymal stromal cell-mediated ex vivo expansion, the expanded cord did not contribute to CD3+ engraftment or long-term hematopoiesis.36
Another potential strategy to achieve more rapid engraftment is to improve homing of HPCs to the recipient’s marrow. Fucosylation of selectin ligands on HPCs is critical for the rolling of HPCs on P- and E-selectins expressed by hematopoietic microvascular endothelium37, 38, 39, which is the initial step of homing to the marrow. Fucosylated HPCs are responsible for engraftment in the NOD scid gamma mouse model40 and PBSC and BM HPCs have >2 × the level of fucosylation of UCB HPCs. Levels of fucosylation can be increased by incubation with α1–3 fucosyltransferase (FT)-VI.38, 40, 42 Robinson et al.41 showed that fucosylation of CB HPCs can be increased from a baseline of ~15% to >95% with a 30-min incubation with FT-VI at room temperature and that only the fucosylated fraction of CD34+ HPCs contributes to engraftment in NOD scid gamma mice. In addition, appearance of human CD45+ cells in the peripheral blood of NOD scid gamma mice was more rapid in mice receiving fucosylated grafts than those receiving untreated grafts and both myeloid and B-lymphoid engraftment in marrow and spleen were threefold greater than in recipients of unfucosylated grafts. Preliminary results of a phase-I study have shown encouraging results with a median time to neutrophil and platelet engraftment of 14 and 33 days, respectively.43 Patients received dUCBT and the cord with a smaller number of total nucleated cells was utilized for the fucosylation procedure. It is unclear from these preliminary results what the effects on long-term hematopoiesis and immune reconstitution will be. However, theoretically, this process could be simply combined with one of the two ex vivo expansion methods described above and this may result in even more rapid engraftment. Cutler et al.44 used an alternative strategy involving the brief ex vivo incubation of one of two CB units with dimethyl prostaglandin E2. Dimethyl prostaglandin E2 has pleotropic effects on HPCs, upregulating genes involved in stem cell homing, proliferation and survival.45, 46 Long-term engraftment occurred from the manipulated unit in 10 of 12 cases, but numbers of patients treated are too small to draw strong conclusions from this observation.44
Grade II–IV and III–IV aGvHD occur in ~30–50% and 10–20% of UCBT recipients, respectively.7, 10, 11, 12, 13, 26, 28, 47, 48 Despite the lower incidence of aGvHD relative to adult donor transplants and the lower incidence of cGvHD relative to all but fully matched sibling transplants, GvHD remains a major cause of morbidity and mortality post UCBT.
The level of CD4+, CD25+ and FOXP3+ regulatory T-cells (Tregs) correlates inversely with aGvHD and cGvHD.49, 50, 51, 52 Although overall Treg numbers and function are preserved, in patients developing aGvHD, there is a marked depletion of naive Tregs which occurs as an early event in aGvHD.53 In cGHVD, there is impaired thymic generation of naive Tregs (CD45RA+) and increased apoptosis of proliferating memory/effector Tregs (CD45RA−).
Treg differentiation and function remains stable even in highly inflammatory conditions,54 which suggests they may be useful as adoptive immunotherapy in aGvHD and cGHVD. However, their in vivo survival in this setting remains to be determined.55 Ex vivo expansion of UCB-derived Tregs may be more attractive than PB-derived Tregs, as the T-cell repertoire is less complex and predominantly naïve, with a well-defined Treg subset present in relatively higher numbers than in PB.56, 57 After flow cytometry-based sorting of CD4+/CD25+ T-cells, UCB-derived Tregs can be expanded ~100-fold using CD3/28-coated microbeads and IL-2. These cells express FOXP3 and maintain potent suppressor function.58, 59 Expansion using IL-15 may be superior. 60, 61 Several murine xenogenic GvHD models have shown that infusion of human UCB-derived Tregs conferred protection from aGvHD and improved survival.62, 63
Although there is concern that Treg infusion post-transplant may potentially impair immune reconstitution, Tregs may suppress alloreactive T-cells, whereas actually facilitating functional immune reconstitution through inhibition of GvHD-induced thymic and secondary lymphoid microenvironmental damage.64, 65, 66, 67 Immune function is preserved in murine models of Treg infusion, suggesting that the suppression of GvHD is relatively specific.68
Another theoretical concern with adoptive transfer of Tregs is the potential for impairment of GVL effect with increased risk of disease relapse.69 Data from mouse models is somewhat contradictory; tumor regression after IL-2-diphtheria toxin-mediated Treg depletion was seen in one model suggesting that Tregs may adversely affect disease-control.70 In contrast, simultaneous adoptive transfer of Tregs and unselected donor T-cells in mismatched mouse transplant models showed protection from GvHD without impairing tumor control.71, 72 It has been suggested that this apparent specificity of Tregs for alloreactive T-cells rather than tumor-reactive T-cells may relate to a different mechanism for the development of GvHD and GVL effect. GvHD may predominantly occur owing to alloantigen-driven expansion of alloreactive T-cells, whereas GVL effect may rely predominantly on activation of alloreactive T-cells.72 Thus, the adoptive transfer of Tregs post-transplant may allow specific inhibition of GvHD, without impairing immune reconstitution and GVL effect.
A preliminary human study has shown post-transplant adoptive transfer of Tregs to be safe, with a reduction in risk of grade II–IV aGvHD relative to historical controls and similar disease-free survival.73 There did not appear to be an excess risk of disease relapse or infection in these patients.
There is delayed neutrophil recovery after UCBT relative to BMT or PBSCT and an increased infection risk which persists beyond the period of neutropenia, with ongoing risk of bacterial, viral74 and fungal infection.75
There are both qualitative and quantitative differences in lymphocyte recovery post-UCBT, with delayed recovery of lymphocyte numbers, particularly naive T-cells, which are present in small numbers until 9–12 months. Fungal infections (particularly invasive aspergillosis) are common, especially prior to D100, but also post-D100.75 Viral infections, especially CMV, adenovirus (ADV), respiratory syncytial virus, varicella-zoster virus, Epstein-Barr virus (EBV)-driven post-transplant lymphoproliferative disease and human herpes virus 6 are seen both pre- and post-D100 and represent a major challenge.75 Patients transplanted with RIC have more rapid reconstitution of T-cell repertoire complexity than after myeloablative conditioning,76 higher T-cell numbers and complexity at different time points and higher numbers of naive T-cells at 12 months after transplantation. In contrast, patients treated with ATG as part of RIC have a delayed T-cell recovery owing to in vivo T-cell depletion of the graft and a substantially higher risk of EBV reactivation and post-transplant lymphoproliferative disease.77
Treatment for viral infection is challenging. Up to one in three deaths after alloSCT may relate to viral infection.78 Pharmacotherapy, where available, is expensive, toxic and often ineffective.79, 80, 81 Consequently, adoptive immunotherapy for treatment and prevention of viral reactivation/infection post-allograft is attractive.
In the adult donor setting, virus-specific T-cells (VSTs) can be generated from the donor, either by culture with modified APCs82, 83, 84, 85, 86, 87 or peptide multimers.88, 89 Rapid isolation strategies such as ‘gamma catch’ can be utilized when VSTs occur at high frequency in the donor’s blood. This strategy has been successfully used in patients with CMV disease or viremia, with a response rate of 83%,90 EBV, with a response rate of 50–70%91, 92 and ADV, with responses in 5 of 6 patients in a small study.93
Adoptive transfer of VSTs from an adult stem cell donor is effective as prophylaxis and treatment for treatment-refractory EBV, CMV and ADV infections.83, 84 However, there are considerable barriers to successful use of adoptive immunotherapy with VSTs post-UCBT; specifically, there are small numbers of T-cells available for manipulation and their phenotype is naïve.85, 86, 87, 94, 95 Hanley et al. have shown that multi-VSTs against EBV, CMV and ADV can be generated from naïve cord blood cells using a complex method of transducing EBV-LCLs transduced with CMVpp65 via an adenoviral vector and these are highly active against virus, despite recognition of non-canonical CMV and EBV epitopes. These VSTs generated from UCB have been shown to be efficacious for treatment of viral infection post-allograft.96 Ex vivo priming of T-cells requires culture with an APC and the addition of IL-7, IL-12 and IL-15. These cytokines decrease antigen concentration threshold, direct T-cells toward a central memory phenotype and influence TH1 and T-cytotoxic type 1 cell differentiation. This process is currently slow and laborious, which will limit its use for urgent treatment, but may allow prophylactic use. Studies to shorten the procedure are in progress and will likely broaden the use of this approach in the sUCBT and dUCBT settings.
3rd party, banked VSTs from adult donors could be used to treat viral reactivation in UCBT patients.97 Suitable lines are available in approximately 90% of patients, can be rapidly identified and made available and induce high response rates in refractory CMV, ADV and Epstein-Barr virus-driven post-transplant lymphoproliferative disease. Despite the theoretical risk of inducing GvHD due to HLA mismatch, no severe cases of de novo acute GvHD were seen.98 Persistence of these cells may be limited in comparison to infusion of VSTs from an HLA-matched stem cell donor owing to the possibility of immunologic rejection.87 Although unclear from existing data, response to subsequent infusions from the same donor may potentially be inhibited because of an anamnestic immune response.
Five VSTs (EBV, CMV, ADV, BK and human herpes virus 6) can now be produced from adult donors after culture with a peptide mix containing immunodominant antigens of the five viruses.99 This method will hopefully be adapted for generation of VSTs from naive, CB units in the future.
Leukemia relapse is the second most-common cause of mortality post-UCBT.26, 28 Disease-free survival is improved and relapse rates are lower after dUCBT for leukemia compared with sUCBT in adults.100 Potential methods of enhancing the GVL effect include the modification of T-cells with chimeric-antigen receptors (CARs) targeting a tumor-associated antigen and adoptive transfer of NK cells.
Autologous CARs have been used with great success in CD19-positive malignancies in the setting of refractory disease.101, 102, 103, 104, 105 Limited experience exists with infusion of CAR-T-cells post allograft. A pilot study of peri-transplant infusion of allogeneic VSTs modified to express a CAR targeting CD19 has shown these to be safe in small numbers of treated patients.106 The study was not powered to assess the efficacy of this approach. Theoretically, the modification of VSTs with a CAR will enhance CAR T-cell persistence owing to stimulation of the cells via the native TCR by virus, whereas the CAR will direct tumor-specific cytotoxicity. This bi-specific approach using virus and CD19+ tumor-specific CTLs is being translated to the UCBT setting.107 The MD Anderson investigators have used the sleeping beauty transposase/transposon system to electroporate a CD19 construct into the UCB of four patients who received those cells following UCBT.108 The CB-CD19-CAR+ T-cells were well tolerated and accrual to the study continues. Kochenderfer et al.109 have also recently demonstrated the feasibility and effectiveness of utilizing donor-derived T-cells for the generation of a CAR T-cell, with a patient with CLL achieving a CR after infusion of donor-derived T-cells transduced with an anti-CD19/28 CAR. These methods are being adapted by several groups for use in the UCBT setting.
Adoptive transfer of NK cells to enhance anti-tumor effect
NK cells can kill cancer cells independently of antigen recognition and are thus attractive for use in adoptive immunotherapy. Allografts containing higher numbers of NK cells have been associated both with reduced disease relapse and lower rates of GvHD.110 Adoptive transfer of NK cells has previously been limited by the small numbers of circulating NK cells (5–15% of the total lymphocytes) and consequently low numbers obtained in an apheresis procedure.111
Methods of NK expansion include:
Overnight culture with IL-2 or IL-15, combined with CD3 depletion of T-cells and infusion the following day. 112
Longer-term culture in IL-2 and IL-15-supplemented media can expand NK cells 60–80 fold in 2 weeks. Nicotinamide increases in vivo survival and cytotoxic activity. 113
Culture with irradiated EBV-LCLs as feeder cells results in expansion 800–1000 fold within 2 weeks. These cells show upregulation of natural cytotoxicity and activating receptors, greater IFN-γ secretion and greater cytotoxicity against K562 and other tumor cell lines compared with resting or overnight IL-2-activated NK cells. 114 The irradiated, gene-modified CML cell line K562 can be used as a feeder cell for ex vivo NK cell proliferation and these cells can be genetically modified to express a range of surface proteins to enhance NK cell expansion and function, for example IL-15 and 4-1BB ligand, CD64 (FCγRI), CD84 (B7-2), CD137L, truncated CD19 and membrane-bound IL-2.115,116,117,118,119,120,121
NK cells can be generated from HSCs, including CB122 and show similar phenotype and activity against leukemic targets to peripheral blood-derived NK cells.123 These can be generated using aAPCs derived from K562 cell lines116, 120 or in feeder cell-free conditions.122 Achieving large-cell production in GMP compliant conditions may be easier with a feeder cell-free system.122
Potential limitations of NK cell adoptive transfer include:
Rapid development of exhaustion after adoptive transfer despite initial high degree of activity. 124
Lymphoyte contamination of the graft. T-cell contamination should be limited to <1–5 × 105/kg, 125, 126 especially in the context of HLA-mismatched allogeneic stem cell transplantation, to minimize the risk of GvHD. This can be achieved by CD3 depletion. 112 Addition of CD56+ selection reduces B-cell contamination to <1%, which also minimizes passenger B lymphocyte-mediated complications such as EBV-driven post-transplant lymphoproliferative disease and acute hemolytic anemia, the latter a risk if minor ABO incompatibility is present. 127 Two-step NK cell enrichment achieves 99% NK cell purity. 128
Extensive ex vivo expansion may reduce in vivo proliferative potential and long-term viability after adoptive transfer. 120
Defects in homing. K562-expanded NK cells show potential defects in CCR7-mediated homing to lymph nodes. 129 Modification of K562 feeder cells to transfer CCR7 to NK cells via trogocytosis transiently increases NK cell CCR7 expression and improves homing to LNs of athymic mice. 130 Culture with irradiated EBV LCL reduces CD62L-mediated homing to marrow and secondary lymphoid tissues, which can be improved by adding nicotinamide to the culture. 113
Cytokine dependence. A rapid decline in NK cell function occurs when IL-2 is removed from medium. 41 In vivo dependence on IL-2 for persistence is greater with ex vivo expanded cells than fresh, activated cells. 131 In vivo use of IL-2 (which can expand NK cell numbers) can lead to severe toxicity and also to Treg expansion which limits NK cell activation. 132 In contrast, IL-15/IL-15Rα complexes promote NK cell activation and enhanced function without the detrimental effects of IL-2. 133, 134 Whether in vivo use of Il-15/IL-15Rα will rescue NK cells from this phenomenon is not known.
NK cell genes
KiRs (killer-Ig-like receptors) are divided into inhibitory and activating subtypes. Inhibitory subtypes bind to HLA-Bw4, HLA-C1 and HLA-C2 group ligands, although many ligands for activating KiRs are unknown.131 KiRs are inherited as haplotypes (KIR-A and KIR-B). KIR-A haplotypes, found in 1:3 adult Caucasians, have one activating receptor, whereas the KIR-B haplotypes have2. Transplantation in AML from a KIR-B haplotype donor is associated with lower relapse rates and better survival.135 Specific donor KIR genes and HLA types are associated with lower rates of relapse in patients receiving 10/10 or 9/10 HLA-matched-unrelated donor allografts. In particular, donor positivity for the KIR2DS1 activating KIR results in a lower risk of relapse and improved survival owing to increased NK cell-mediated GVL effect. However, this beneficial effect is lost in donors who are homozygous for HLA-C2 as high levels of HLA-C2, which is recognized by KIR2DS1, result in NK cell tolerization. The lowest relapse rates were seen in recipients of a transplant from an HLA-C mismatched donor who was positive for HLA-C1 and KIR2DS1.136 Recipients who are HLA-C2/C2 homozygous are also at higher risk of relapse, but their relapse risk is lower when receiving an HLA-C mismatched transplant from a KIR2DS1 positive donor.136 These observations raise the prospect that, in the setting of AML, if ex vivo NK cell expansion from cord blood is performed, selection of cord units with specific KIR and HLA-C genotypes may be beneficial. Whether the addition of immunomodulatory drugs such as lenalidomide or checkpoint inhibitors such as PD1-antibodies will further enhance NK cell anti-tumor function will be an important future question.
Summary: potential future directions
Given the array of potential cellular therapies that can now be generated from a cord unit, and the limited number of cells available, we are faced with the prospect of having to choose which of these is the most important in an individual patient. For example, a patient deemed to be at high risk of relapse may have part of a cord utilized for NK cell expansion, whereas a patient at high risk for infection could be given prophylactic VSTs post transplantation. This limitation could potentially be overcome by the use of ‘third party’ cellular therapy products for certain applications. For example, third party VSTs could be used when required to treat refractory viral infections rather than infusing CB-derived VSTs prophylactically. This would allow generation of Tregs to minimize GvHD and/or NK cells to maximize anti-tumor effect from the non-expanded cord unit. This cord could be chosen based on its HLA-C/KIR genotype to maximize NK-cell-mediated anti-tumor effect. Another potential avenue to explore in the future would be the use of a third cord to manufacture Tregs and NK cells for adoptive therapy. Alternatively, if two cords are to be used, the first could be split in two, using 50–60% for ex vivo expansion/fucosylation to provide rapid engraftment and the remainder for the generation of Tregs and NK cells to be infused peri-transplant. The other cord, which will provide long-term engraftment, if of sufficient size, could have 20% utilized for the generation of VSTs ±CARs, whereas the remainder is infused unmanipulated to provide long-term hematopoiesis and immune reconstitution.
Clearly, these are exciting times in the field and systematic investigation will be required to determine how best to utilize available resources to maximize the anti-tumor effect and minimize the AE associated with UCBT.
About this article
Biology of Blood and Marrow Transplantation (2017)